Model of pedestrian flows against empirical pedestrian counts for New York City, constructed from “flow layers” formed from pair-wise matching between the following seven categories of origins and destinations:
- subway
- residential
- transportation
- sustenance
- entertainment
- education
- healthcare
An eighth category is network centrality, with additional layers modelling dispersal from each of these categories. The model explains R2= 83.9 of the observed variation in pedestrian counts. Final results, with significantly explanatory layers named according to the first three letters of the above categories, looks like this:
Layer Name | Estimate | Std. Error | t value | Pr(>t) |
---|---|---|---|---|
edu-tra | 23977 | 4484 | 5.35 | 0.0000 |
edu-sus | 16904 | 5572 | 3.03 | 0.0031 |
edu-dis | -78057 | 24521 | -3.18 | 0.0020 |
edu-hea | -24921 | 4445 | -5.61 | 0.0000 |
ent-tra | 38179 | 12019 | 3.18 | 0.0020 |
hea-dis | 105658 | 10706 | 9.87 | 0.0000 |
sub-dis | 23 | 3 | 8.99 | 0.0000 |
sub-hea | 8 | 1 | 6.66 | 0.0000 |
sub-tra | 6 | 1 | 5.08 | 0.0000 |
sub-cen | -10 | 1 | -6.99 | 0.0000 |
sus-res | 6258 | 1232 | 5.08 | 0.0000 |
sus-ent | 1446 | 361 | 4.00 | 0.0001 |
sus-sub | -1337 | 331 | -4.04 | 0.0001 |
sus-edu | -5924 | 978 | -6.06 | 0.0000 |
Table 1. Statistical parameters of final model of pedestrian flows through New York City.
A sample of actual flows looks like this:
And a final statistical relationship between modelled and observed pedestrian counts looks like this: